In cities across the United States this year, unemployment continues to rise, budgets continue to get smaller, and governments find themselves confronting 21st-century problems with 20th-century tools.

Hoping to use technology to change the way city halls confront issues like education or crime, an organization called Code for America launched in late 2010. Through Code for America, each of four cities — Boston, Philadelphia, Seattle, and Washington, D.C. — offered to pony up $225,000 for a team of five programmers, designers and other technologists to spend the year developing specific solutions for their problems.

When the project launched, I saw it as a test of a simple premise: Given coding talent and information-technology knowledge, big municipal governments can make their cities better without spending a whole lot of money. Another high-level objective that I didn't pick up on right away was to bring the rapid-prototyping style of modern technology companies to city government. And there's always the idea that the cool stuff these technologists make will in turn make cities easier to use and more fun to live in.

Turns out this is all harder than it sounds — except for the bit about cool stuff that makes cities more fun, but more on that later. The program last week celebrated a transition from its first year to its second at a conference in San Francisco called the Code for America Summit, where city officials and Code for America fellows described months of work that put in stark relief the differences in culture and approach between local governments and technology startups.

People affiliated with the program sought to describe its first year, now with just a few weeks left, as the beginning of a cultural shift in the way governments work. In some ways, that's empirically true. In Boston, for example, the city will be hiring a full-time "civic coder" for the first time ever — something city Office of New Urban Mechanics co-chair Nigel Jacob says is a direct result of what Boston officials learned through the Code for America fellows' appointment. Seattle Chief Information Officer Bill Schrier said that after their engagement with CfA, the city will start from scratch on the question of "how are we interacting with our community." And included in the software yield from Code for America are some applications that give cities a level of information management ability they never had before. For instance, DataCouch, developed by Boston CfA fellow Max Ogden, makes it easier and faster to share and improve sets of data that other developers use to write applications. Another Code for America team has produced software called the Open311 Dashboard, a visual web interface for non-emergency issue handling for cities that use Open311 standard. A list of largely completed projects is here.

But the output of Code for America's first year also showed that startup-oriented techies and dyed-in-the-wool public servants are often very far apart in their thinking. The premise of the entire project was to see what would happen if programmers accustomed to the latest private-sector tools — languages like Python or frameworks like Ruby on Rails, new development styles with names like "agile" or "scrum," offices paid for with venture capitalist money and pantries full of cereal and refrigerators full of beer — found themselves suddenly in the public sector. Many anticipated a certain degree of culture clash. The program's directors tried to prepare the technologists, who would spend the year on $35,000 fellowships, with an entire curriculum on how cities work. Some city hall staffers, like Philadelphia's Desiree Peterkin Bell, described last week how they planned meetings and briefings for top officials in order seek acceptance for the Code for America program.

"We just had to get them out in front of as many people in city government as possible," Bell's colleague, Jeff Friedman, said Friday.

By doing that, Friedman says, the city of Philadelphia gave their fellows room to work.

"You guys showed me a whole new way of doing things," said Friedman, Philadelphia's manager of civic innovation and participation. "We have a lot of rules, we're constrained by how we can procure and hire and manage projects. You guys were like a flock of gazelles just running across the plain, doing your thing. It was kind of inspiring."

But this idea was so obviously going to lead to confrontations around culture and organizational structure that the headline of an article last year from Fast Company's Anya Kamenetz included Code for America not as a graceful herd of herbivores but instead as "an army of techies" that was "taking on city hall." Turns out that invasion met heavy resistance on many fronts, starting with their jeans-and-sneakers uniforms.

Max Ogden. Photo: Isaac Schlueter / Flickr

"We should have told them to wear ties," Jen Pahlka, Code for America's founder, explained Friday.

In some ways, Pahlka and some Code for America fellows described to me a colossal mismatch of expectations. After announcing that CfA would parter with the government of Washington, D.C., the administration there changed — and the chief technology officer they had planned to work with was replaced. D.C. was taken off the roster of participating governments. In Boston, the city had proposed that the fellows build an application to provide a unified location for educational data — only to be thwarted along the way. Last year, Code for America chief geek Dan Melton announced that the fellows would be linking student data to a universal student ID card that is already in the pilot phase in Boston. That didn't happen.

"They tried to do a data platform, but couldn't get the data," Pahlka explained.

Max Ogden, a Code for America fellow who has a reputation as something of a magician with the newfangled database software CouchDB — and, with hair past his shoulders and a forked red beard that goes down to his chest, really does look like a wizard — gave me one example of culture clash. He was supposed to head to a meeting with some lawyers to discuss the Code for America project, only to have the meeting canceled. Instead, the lawyers dealt with school closures.

There were practically people with pitchforks outside, Ogden told me, and there he was, talking about open data.

For Philadelphia and Seattle, the fellows were tasked with providing an application for civic leaders to find each other and organize around projects that anyone could propose. Schrier, though, described deep disagreement with the Code for America team on how that project should have been managed.

"The big clash of culture that we had was actually around project management," Schrier said. "We have a very, some people would say, even draconian project management model in Seattle for managing IT projects and of course something like Code is pretty antithetical to that. Matching those two cultures up was very difficult."

The city was expecting a new software product built according to Seattle's way of doing things. What they got instead was a software platform, Change By Us, originally developed by the New York-based firm Local Projects for the nonprofit CEOs for Cities. CfA's contribution was twofold. First, they worked with Local Projects to take the program's proprietary code and make it open-source. Second, they're continuing to work on expanding its capabilities — this Code for America year still has a few weeks left.

Change By Us, originally launched in New York and based on an earlier piece of software called Give a Minute, gives neighborhood activists a way to connect to one another around given topics or projects. CfA fellow Chacha Sikes also created another application, CityGroups, to fulfill Seattle's desire for a way for people in neighborhoods to find traditional groups like block associations, and for neighborhood watch leaders to connect with one another around crime issues. All of this is mostly built and already up and running in Seattle — but, to use startup-speak, the team made a "pivot" to this plan over the summer, after Seattle officials had developed different expectations.

Another group of fellows, with project manager Matt Lewis, started work on "JobOps," a web application for veterans returning to civilian life. Previewed at last week's event, JobOps takes a veteran's military occupation code — a classification all members of the armed forces receive — and uses that to dig up data on where that veteran might start looking for a related career in civilian life, complete with job descriptions, suggested additional credentials to get, like an emergency medical technician qualification or a professional certificate, and potentially relevant groups on social networks like LinkedIn.

JobOps isn't done yet; on Friday, Lewis had screenshots replete with lorem ipsum and a demo video that wouldn't play through his presentation software. Again, his team has a few weeks to finish up. But Pahlka is clearly not interested in working with the federal government next year. (The State Department applied to participate in Code for America and didn't wind up involved for 2012.) When pressed, Pahlka told me that the paperwork involved in working with a federal agency is just too much for her currently small staff to handle.

"There's good service," Clay Johnson said Thursday at the Code for America Summit, "there's mediocre service, there's 50 feet of crap, and then there's government service." Photo: Ryan Resella / Flickr

Given all of this, Code for America supporters credit the 20 fellows who worked around the country this year as much for giving this whole run-government-like-a-startup thing a shot as anything else. After a year of trying, it's clear to Pahlka and many people associated with the program that if technology can help to solve the problems that local governments have, that won't get done in just a handful of months.

Clay Johnson, the former director of Sunlight Labs, a co-founder of the political software firm Blue State Digital and a member of Code for America's board, on Thursday likened their experience to Billy Beane's time with the Oakland Athletics as portrayed in the film Moneyball. Beane's willingness to take a data-driven approach to baseball and reject some of the magical thinking that goes into managing a baseball team made him a controversial figure. After over a decade of leaning on data to identify undervalued players they could afford with a meager baseball budget, the Athletics would finally win the 2006 American League Division Series but lose in the conference championships.

Theo Epstein, the general manager of the Boston Red Sox and a fellow proponent of sabermetrics, used some of the same thinking en route to the better-funded Sox's victory over the Yankees in their curse-breaking World Series of 2004. On opening day 2004, the Sox also had the second-highest payroll in baseball — behind the Yankees — but that's for another kind of blog. The point of Johnson's argument is that asking traditional institutions to change how they work — whether that's Fenway Park or Boston City Hall — is a multi-year process that begins with controversy, tension, and pain.

"The amazing thing that the fellows did and the cities did this year was they were the first ones through the wall," Johnson said, before a crowd packing an event space on the second floor of a four-floor building not far from the Yerba Buena Arts Center in San Francisco. "They took the hits so you don't have to."

Johnson, known for having a flair for the dramatic (such as the profane tumblr linked to here), may have been overstating CfA's underperformance. Last week, Pahlka couldn't stop talking about DiscoverBPS, another one of the applications that the Boston team built instead of what they had originally set out to create. Demonstrated at Code for America Summit, DiscoverBPS is meant to help parents find schools close enough to their homes that their children will get priority enrollment. Parents put in their address and should be presented with a map of schools inside a given radius. Things are a little more complicated than that, though — the site allows parents to rank schools and flag them to come back to later. It also uses detailed city data so that it can count a school as within walking distance if part of the parcel of land its on falls inside the required radius, rather than just using its geocoded address.

"A site of that quality with the few resources that went into it, that's not an outcome that's currently considered possible in government," said Pahlka, who managed the video game industry-covering Gamasutra and event production for events like Game Developers Conference and the Gov 2.0 events during stints at CMP Media and TechWeb.

As she pointed out, Discover BPS is just one example. The project that started out as something in partnership with Washington, D.C., instead turned into a Civic Commons, quasi-nonprofit in its own right, which is co-incubated with New York-based OpenPlans and dedicated to fostering the use and sharing of open-source software by governments. Former White House Deputy Chief Technology Officer Andrew McLaughlin is its executive director.

The lines of working code from the past year also stretch pretty far, and people seem to have noticed. Within the first two hours of last week's event, two people from two different companies stood up and announced job openings for Code for America fellows.

I spoke to several of the Code for America fellows about what they planned to do, and while all of them were interested in civic life, none were ready to march back to a city hall somewhere and get back to work. That highlights another one of the challenges to Code for America's basic premise. This year saw a marked brain drain from government, with several high-ranking federal IT officials leaving and commenting on the glacial pace of change on their way out.

This is part of the problem Code for America was built to solve.

"In the world," Johnson said during his Thursday talk, paraphrasing a quote from Beane's character in Moneyball, who was speaking of the Oakland A's, "there's good service, there's mediocre service, there's 50 feet of crap, and then there's government service."

But shoveling cities out from under that proverbial pile is a lot to ask of what Pahlka hopes will be top-tier technologists, some of whom could just as easily get six-figure salaries at big-name software companies. Pahlka says it will take more than another year to arrive at a way to continue to grow their organization in a sustainable way.

"We need at least three years to figure out what we really are," she said of Code for America.

Already, though, she says that the organization has at least proved its premise — from culture clash has come cultural change, and with it the idea that maybe a cash-strapped city doesn't need to pay millions of dollars just to get a decent website.

"We're changing perceptions inside and outside government," Pahlka said, "about what government can do and at what cost."

Code for America is also changing, asking cities to take fellows on who will work to address issue areas but not to accomplish specific tasks as they did last year. Instead of five teams in four cities at $225,000, fellows will work in eight teams of three of a cost to each city of $150,000. Many cities will have their way paid for them by foundations or other grant money. And preparation for fellows and cities started earlier for 2012, with a meeting for 2012 cities held last week. This all makes Pahlka think the idea is catching on.

"There were so few cities when we started who wanted to partner with us," she said. "And now, there's a lot more."